MSc. in Software Technology Dissertation of Sender-Adaptive Rate Control for Layered Video Multicast

This dissertation proposes the Sender-Adaptive Rate Control for Layered Video Multicast[SARC] to provide a rate control mechanism for the layered video multicast over the internet. Unlike other established approach, SARC will determine at the start of the video session the number of video layers according to the real-time status of the network bandwidth and make adaptation to the video layers so that the convergence time would be minimal. SARC adopts the hierarchical structure from LVMR[10] to organize the receivers into domains. Intermediate Agent/IA administers the domain and gathers network status information from different subnets within the domain. Subnet Agents/SA are responsible for their own subnets. In SARC the sender gets the real-time network status information from Intermediate Agent. It consolidates various link speeds of the receivers. It then dynamically splits or combines video layers as well as the transmission rate of each layer so that the convergence time would be minimal. The SARC is evaluated by simulation using network simulator ns. The result is compared with established approach such as RLM[7]. In RLM[7] the sender is not involved in the rate control mechanism, only receivers are involved. SARC is shown to have faster convergence time and lower loss ratio than RLM[7].